The complete working record of a personal research program: the bet that physics is, at bottom, the geometry of a manifold — and the honest account of how far that bet got.
Status: the record is closed at phase 103 (2026-07-19). The framework's deepest gate — the join — was answered in the negative for everything constructible, and a pre-registered kill fired. What survives is stated in §"What stands" below.
Over 103 numbered working phases the program pursued a single thesis: that a smooth 3-manifold Σ (with time emergent as modular flow) is ontologically primitive, and that forces, quantum structure, matter, and the arrow of time are derived from that geometry rather than added to it. The organizing discovery was that geometry fixes physics only on a protected stratum, and the central conjecture (Plank IV) was that the protecting structure is the stiffness of spacetime — the Godbillon–Vey invariant of a codimension-1 foliation — whose modular flow is the clock.
The program was run under standing honesty rules, and those rules are the reason this repository is worth reading even where the physics failed:
- Every claim carries a tag: [proved] / [established (others')] / [conjecture] / [interpretation] / [fitted].
- Predictions were pre-registered with kill conditions before the data.
- Negative results are reported plainly, never buried or tuned away.
- Kill tests were armed in advance and honored when they fired.
- No fabricated numbers, assignments, or derivations — compute, cite, or say "unknown."
Four red-team campaigns (two internal, two context-free adversarial) audited the record; their reports are preserved verbatim.
| Read this | For |
|---|---|
core/CORE-geometry-first-v7.md |
the authoritative core document (7th issue; supersedes v6 and all earlier CORE-*) |
status/STATUS-plain-english-2026-07-17.md |
the program in ordinary language |
essays/ESSAY-the-fully-wound-world.md / essays/fully-wound-world.html |
the narrative essay version |
status/STATUS-claims-novelty-publishability-2026-07-18.md |
what can honestly be claimed, what is novel, what is publishable |
notes/QUESTIONS-2026-07-19.md |
the complete open-problem ledger — who could answer each, and what dies if the answer is wrong |
planning/PREDICTIONS-LEDGER-2026-07-17.md |
the falsifiable commitments, with referees and dates |
planning/research-map-MASTER.md + planning/research-map-ADDENDUM-2026-07-17.md |
phase-by-phase index of all 103 phases |
core/CORE-geometry-first-FINAL.md/.pdf, core/CORE-geometry-first-and-the-protected-stratum.md/.pdf, and core/CORE-final-v2.md are superseded; they are kept for provenance only.
Mathematics, refereed and independently verified.
- DTAB — the Heegaard Floer d-invariants of circle bundles over surfaces (
data/DTAB-PAPER.md, data indata/DTAB-DATA-2026-07-19.md, computations inscripts/dtab*_calc.py). A standalone mathematics paper with no framework content: the graded 𝔽[U]-module structure of HF⁺(Y(g,n), 𝔰) for every torsion spin^c structure. The program's most credible artifact — printed proofs, runnable scripts, checkable in an afternoon. - The base-inertness theorem and the constructive census (
phases/phase103-JOIN3c-census-closed.md): every canonical modular flow and Wiesbrock/Borchers translation from leg-aligned subalgebras fixes the base pointwise. This is the theorem that closed the program's central hope — and it is the cleanest thing the join chapter produced. - NOTE-1 (
notes/NOTE-1-cpt-commutant.md): the Boyle–Turok involution as modular conjugation of a constructed standard form. Two independent referees searched for prior art and found none. notes/REMARK-1-circle-bundles.md,notes/NOTE-2-unique-KMS.md— smaller results, correctly attributed.
Negative results, which are results.
- The κ go/no-go: NO-GO (
phases/phase38-tier2-2B-kappa-go-no-go.md). Sorkin triple-path interference κ = C·gτ/ħ computed explicitly at leading order; all three Godbillon–Vey enhancement routes closed. κ is Planck-suppressed and retired as a fingerprint. - The AMK predictivity audit: negative (
phases/phase42-T2-verdict.md). The exotic-smoothness numerical channel is retrodictive fitting dressed in rigid invariants — no shared continuous modulus exists to eliminate. - The Hodge-weight obstruction, the global-hyperbolicity divide — theorems, not bugs; stated rather than fixed.
Live cosmological commitments (planning/PREDICTIONS-LEDGER-2026-07-17.md), registered with timestamps and kill conditions before the data: rigid dark energy (w₀, wₐ) = (−1, 0), with a pre-filed successor if it evolves. Referee: the sky.
The framework as physics. Every external referee located the reason identically: the dictionary — the functor joining the geometric layer to the observable net — does not exist. The named gaps (EQ, the join, F-GAP, T-2π, A1) are precisely where the physics would have been. The geometry did not become physics, and the record says so in its own last line.
The archive is organized into folders by document type:
| Folder | Contents |
|---|---|
core/ |
CORE-* — the core document across seven issues (v7 authoritative), plus superseded issues kept for provenance |
phases/ |
phase1–phase103-*.md — the working record, in order |
reviews/ |
REPORTS-fresh-red-team-*.md (the adversarial referee campaigns, verbatim) and REVIEW-*.md |
notes/ |
NOTE-*, REMARK-* standalone mathematics; plus QUESTIONS-*, CODA-*, torsion-U1-* |
data/ |
DTAB-* — the d-invariant paper and its data |
essays/ |
ESSAY-*, INTERPRETATIONS-*, fully-wound-world.html — narrative/plain-language versions |
outreach/ |
PACKET-outreach-v*.md, SIGNATURE-block-template.md — correspondence infrastructure |
scripts/ |
*_calc.py — every computation, reproducible |
papers/ |
reference index (papers/README.md) with full citations and arXiv/DOI links to the source papers cited throughout — the papers themselves are linked, not redistributed |
status/ |
STATUS-* — status and plain-English summaries |
planning/ |
TIER-2-KICKOFF.md, TIER-3-PLAN.md, RESEARCH-PLAN.md, research-map-*, PREDICTIONS-LEDGER-* — the plans (all executed) and the phase index |
The Python scripts are self-contained and depend only on the standard library plus sympy/numpy where noted:
python3 scripts/i3_calc.py # the leading-order Sorkin I₃ computation (phase 38)
python3 scripts/dtab4b_calc.py # the d-invariant sweep engine (DTAB paper)PDFs are built from the Markdown with pandoc + xelatex:
pandoc IN.md -o OUT.pdf --pdf-engine=xelatex -V geometry:margin=1in \
-V fontsize=11pt -V colorlinks=true --tocThis is an open research record and collaboration is welcome — corrections, new calculations, referee-style critiques, and attacks on the open problems in notes/QUESTIONS-2026-07-19.md especially. Before opening an issue or PR, please read CONTRIBUTING.md: the program runs under standing honesty rules (claim tags, pre-registered kill conditions, no fabricated numbers, negative results reported plainly), and contributions are held to the same bar. Open-ended discussion belongs in Discussions; specific problems and proposed fixes in Issues.
This repository is dual-licensed:
- Written content — all Markdown documents, notes, essays, data tables, and the
.pdfwrite-ups authored here — is licensed under Creative Commons Attribution 4.0 International (CC-BY-4.0). You may share and adapt it, including commercially, with attribution. - Code — everything under
scripts/— is licensed under the MIT License.
Third-party source papers are not redistributed here; papers/ contains only a citation index with links to the originals, which remain the property of their respective authors and publishers and are not covered by the licenses above. To cite this work, see CITATION.cff.
It is not a claim to have found a theory of everything. It is a complete, timestamped, adversarially audited record of a serious attempt that failed at a specific, identifiable place — plus the handful of theorems that are true whether or not the universe agrees. The method, not the conclusion, is the part offered for use.